2.1.   factorial.c

/*
 * This file is part of mutest, a simple micro unit testing framework for C.
 *
 * mutest was written by Leandro Lucarella <llucax@gmail.com> and is released
 * under the BOLA license, please see the LICENSE file or visit:
 * http://blitiri.com.ar/p/bola/
 *
 * This is an example module that calculates a factorial.
 *
 * Please, read the README file for more details.
 */

unsigned factorial(unsigned x) {
	if (x <= 1)
		return 1;
	return x * factorial(x-1);
}

2.2.   factorial_test.c

/*
 * This file is part of mutest, a simple micro unit testing framework for C.
 *
 * mutest was written by Leandro Lucarella <llucax@gmail.com> and is released
 * under the BOLA license, please see the LICENSE file or visit:
 * http://blitiri.com.ar/p/bola/
 *
 * This is the factorial module test suite. Each (public) function starting
 * with mu_test will be picked up by mkmutest as a test case.
 *
 * Please, read the README file for more details.
 */

#include "factorial.h"

#include "../mutest.h"

void mu_test_factorial_zero() {
	unsigned x = factorial(0);
	mu_check(x == 1);
}

void mu_test_factorial_one() {
	unsigned x = factorial(1);
	/* this test is wrong on purpose, to see how it fails */
	mu_check(x == 2);
}

void mu_test_factorial_positive() {
	unsigned x = factorial(2);
	/* this test is wrong on purpose, to see how it fails */
	mu_check(x == 3);

	x = factorial(3);
	/* we don't want to continue if this fails, because the next result
	 * depends on this one. This one will succeed. */
	mu_ensure(x == 6);

	x = factorial(x);
	mu_check(x == 720);

	x = factorial(4);
	mu_ensure(x == 6); /* same as before, but this one will fail. */

	x = factorial(x-15); /* and this will never be executed */
	mu_check(x == 362881); /* but if excecuted, will fail */
}

2.3.   exception_test.cpp

/*
 * This file is part of mutest, a simple micro unit testing framework for C.
 *
 * mutest was written by Leandro Lucarella <llucax@gmail.com> and is released
 * under the BOLA license, please see the LICENSE file or visit:
 * http://blitiri.com.ar/p/bola/
 *
 * This is a C++ module test suite. It shows how to use checks involving
 * exceptions.
 *
 * Please, read the README file for more details.
 */

#include <stdexcept> // std::out_of_range
#include <vector> // std::vector

#include "../mutest.h"

extern "C" {

void mu_test_exceptions() {
	std::vector<int> v(1);
	// ok
	mu_check(v.at(0) == 0);
	// throws! This fails
	mu_check(v.at(1) == 0);
	// ok, we expect the exception to be thrown, and it does
	mu_echeck(std::out_of_range, v.at(1));
	// fails! We expect this to throw, but it doesn't
	mu_echeck(std::out_of_range, v.at(0));
	// fails again, but this time the show is over (note the "ensure")
	mu_eensure(std::out_of_range, v.at(0));
	// this will never be executed (it should fail if it is)
	mu_check(v.empty());
}

} // extern "C"

3.1.   Test Program

A test program is the higher level unit of mutest. The test program is the one in charge of running all your tests. Probably one of the more important features of mutest is that you are not supposed to bother about the test program. So, different implementations have different ways to tackle this. Some need more or less interactions from your part, and each have their pros and cons.

But this is all you need to know for now, for more details see how the test program is implemented by your implementation of choice.

3.2.   Test Suite

A test suite is the higher level unit of mutest that you should care about =). Is not much more than a way to group test cases. Code-wise, a test suite is a C (or C++) module (or compilation unit). Not clear enough? A unit test is an object file (could be a shared object depending on the implementation). This module should have one or more test cases and it could have any number (including zero) of initialization and termination functions.

A test suite, is inspected by the test program for test cases and initialization and termination functions, and run them.

A test suite fail if one or more test cases fail, and it's skipped if one or more initialization functions fail (or, depending on the implementation, if the test suite can't be loaded at all).

3.3.   Test Case

A test case is just a plain function with a special signature and name. A test case function name must start with mu_test, and take no arguments and return nothing. For example:

void mu_test_something(void);

A test case (probably) only make sense if it has checks. A test case succeed only if all its checks succeed too.

Test are executed in an implementation-dependant order, but usually the default order is alphabetical.

3.4.   Checks

Checks are assertions that a test case must pass (a boolean expression that must evaluate to true). There are 2 big flavors of checks: check and ensure. check just print an error (and mark the test case as failed) and ensure halt the test case execution, jumping to the next one.

For better C++ support there are check macros that assert that a specified exception is thrown (instead of check for a boolean expression to evaluate to true).

You can take a look at the reference to see the different flavors of check macros in more detail.

3.5.   Initialization

Sometimes you need to setup some environment shared between all the test cases in a test suite. You can use initialization functions for this.

An initialization function, like a test case, is a plain C function with a special name and signature. The name must start with mu_init and it must take no arguments, and return an error code (0 being success). For example:

int mu_init_something(void);

All initialization functions are executed before any test case, in an implementation-dependant order, and if one of them fail (returns non-zero), the whole test suite is skipped immediately.

3.6.   Termination

Termination functions are just like initialization functions, but they're executed after the test cases, their names start with mu_term and they return nothing. For example:

void mu_term_something(void);

5.1.   Static Implementations

Static implementations can be only written in C/C++ (or other language that is link-compatible with C, like the D Programming Language, but since one of the main goals of mutest is avoid unnecessary dependencies, you probably don't want to depend on an extra language/compiler to run your tests =).

The main advantage is better debugging support, because you can run the test program in a standard debugger and see what happens with test cases very naturally.

The main disadvantage is, the test suites must be figured out in compile-time, usually using some kind of code generation (if you want to avoid writing repetitive code yourself). There's also a limitation in the test case, initialization and termination functions names: they should be unique for all the test program.

$ cc -c -o mutest.o mutest.c
$ cc -c -o test1.o test1.c
$ cc -c -o test2.o test2.c
$ mkmutest mutest.h test1.o test2.o | cc -xc -c -o runmutest.o -
$ cc -o testprg mutest.o test1.o test2.o runmutest.o

$ ./testprg

5.2.   Dynamic Implementations

Dynamic implementations, on the other hand, can be written in any language that can access to shared objects. The idea is to inspect a shared object for test suites and run them, without requiring any information about test suites at compile time.

There are several advantages in this kind of implementations. The dynamic nature let you completely separate the test program from the user-written test suites and you can choose at run-time what test suites to execute by just selecting the correct shared objects. Also, test case, initialization and termination functions names only have to be unique in the scope of the test suites, because test suites are completely isolated in separate shared objects.

But everything comes at a price, and the higher price to pay is debuggability. It's a little harder to plug a debugger to a shared object.

$ cc -c -fPIC -DMUTEST_PY -o test1.o test1.c
$ cc -shared -o test1.so test1.o
$ cc -c -fPIC -DMUTEST_PY -o test2.o test2.c
$ cc -shared -o test2.so test2.o
$ mutest test1.so test2.so

6.1.   mu_check()

Synopsis

mu_check(expression)

Description

Check that the expression evaluates to true. Continue with the test case if fail.

Availability

Always

Example

void mu_test(void)
{
    mu_check(5 == 4); /* fail */
    mu_check(5 == 5); /* excecuted, pass */
}

6.2.   mu_ensure()

Synopsis

mu_ensure(expression)

Description

Check that the expression evaluates to true. Interrupt the test case if fail.

Availability

Always

Example

void mu_test(void)
{
    mu_ensure(5 == 4); /* fail */
    mu_check(5 == 5); /* not excecuted */
}

6.3.   mu_echeck()

Synopsis

mu_echeck(class, expression)

Description

Check that the expression throws a specific exception class (or subclass). Continue with the test case if fail.

Availability

C++ only

Example

#include <stdexcept>

extern "C"
{
    void mu_test(void)
    {
        mu_echeck(std::exception, true); /* fail */
        mu_echeck(std::exception,
                throw std::runtime_error("!")); /* excecuted, pass */
    }
}

6.4.   mu_eensure()

Synopsis

mu_eensure(class, expression)

Description

Check that the expression throws a specific exception class (or subclass). Interrupt the test case if fail.

Availability

C++ only

Example

#include <stdexcept>

extern "C"
{
    void mu_test(void)
    {
        mu_eensure(std::exception, true); /* fail */
        mu_echeck(std::exception,
                throw std::runtime_error("!")); /* not excecuted */
    }
}